Charting Better Routes to Drugs

At an invitation-only symposium with an unusual workshoplike format, researchers from the pharmaceutical industry, academia, and government gathered last month for a series of unusually frank discussions and debates on a set of provocative and controversial drug discovery issues: Why haven't chemists embraced "grand challenge" scientific initiatives the way biologists have? What sort of chemistry projects of that kind might be worth pursuing? Have recent drug industry investments in advanced technologies been wasteful and unproductive? Should researchers in academia try to discover new drugs? Are drug company hiring practices rigid and counterproductive? And are pharmaceutical companies running out of biological targets to "hit" with new drugs?

The symposium, "Charting Chemical Space: Finding New Tools To Explore Biology"--part of the twice yearly Horizon Symposium series cosponsored by Aventis and Nature Publishing Group--was held at a scenic inn on the coast of Maine.

Chemical space is the universe of all possible molecular structures, and the meeting on this topic was the fourth Horizon Symposium and the first to focus to a large extent on chemistry.

Horizon meetings are unusual in that they encourage open discussion among all attendees and greatly de-emphasize formal scientific presentations, restricting them to 10 minutes each. The symposia are designed as "an antidote to most meetings we go to, where formal talks are the norm and interesting discussions an afterthought," said Adam Smith, editor of Nature Reviews Drug Discovery, in opening remarks at the meeting.

"It was a very ambitious meeting to try and address very complex issues, bringing together perspectives on the range of chemistry relevant to biology and issues involving academia and the drug industry," said professor of chemical and structural biology Christopher M. Dobson of Cambridge University, who served as moderator for the symposium's opening and closing sessions. "As with many meetings, it got stuck at some points, but somehow it then all loosened up, and I think nearly everybody felt they learned a lot about some very important issues and went home excited about the possibilities for the future."

"All meetings can't be like this," said professor of chemistry and molecular biology Gerald F. Joyce of Scripps Research Institute. "Gordon Conferences aim to be like this, but people there still get up and tell their own story. Here, you really are obliged as a presenter to digest what the story is for the field and communicate it to your colleagues."

Chemistry professor Barry M. Trost of Stanford University agreed that this is the kind of meeting where you're trying "to evaluate where a field is, and that really means having more discussion on the issues. The meeting certainly did that to a much greater extent than I have seen at any other meeting."

The success of such a meeting, however, "depends on what people you get together--both presenters and attendees," Trost said. For instance, "some session leaders were very good at raising talking points in their presentations," whereas others reverted to more traditional presentations that didn't encourage discussion. Meeting organizers tried to discourage the latter from the outset "but did not always succeed in meeting their laudable goal," he said. "So the format is a useful one, but I think the organizers need to work on it a little bit more."

In any case, "I really do think this is the way a meeting should be done," commented Aseem Z. Ansari, an assistant professor of biochemistry at the University of Wisconsin, Madison. "I've never encountered big pharma people expressing themselves publicly in the way they did here--with quite a lot of candor and honesty. With the group assembled at this meeting, the format worked."

GRAND CHALLENGES. A key topic at the meeting was the concept of grand challenges for chemistry. Attendees like Dobson and Stuart L. Schreiber, a Howard Hughes Medical Institute investigator and professor of chemistry and chemical biology at Harvard University, noted that biologists have organized a number of grand challenge initiatives and resources--projects to which many researchers contribute and from which many can benefit--but that chemists for the most part have not done so.

Examples of grand challenge biology projects include the Human Genome Project, the Single Nucleotide Polymorphism Consortium, and databases like GenBank (an annotated collection of publicly available DNA sequences) and the Protein Data Bank (a repository of biological macromolecule 3-D structure data). "With all the technologies coming out for searching databases and the power that's had in biology," Dobson said, "it would be tremendous to set up comparable databases for chemistry. And I think it is important, if one wants academics to bring their ideas in, that it should be in the public domain as much as possible."

Researchers have seen how well the grand challenge culture can work in biology, Schreiber said. So why can't chemists behave like this, too, he asked--by sharing their data in an open format and moving toward common standards for such shared resources? He proposed two chemistry grand challenge ideas: building a comprehensive database of bioactive compounds and associated bioactivity screening data, and discovering a small-molecule modulator for each individual function of every human protein.

Schreiber noted that PubChem, a National Institutes of Health small-molecule and assay data repository, is in the planning stage; that ChemBank, his own group's small-molecule and assay data analysis database, is already operational; and that both are (or will be) free and openly accessible resources. The grand challenge database of bioactive compounds he proposed would complement such resources and perhaps be more comprehensive. He also noted that a grand challenge database would be free, distinguishing it from current databases such as the Beilstein file and the Chemical Abstracts Service Registry. "Free access helps catalyze use of a resource and stimulates insights," Schreiber said.

"A new central chemical information repository could take [bioactivity] data that has been generated with public funding and make it readily available to the community," Ansari said. "What I gathered from discussions at the meeting is that such data in chemistry [currently] are not accessible, not shared, and are far more proprietary than in biology, and that has to change. The symposium will influence people to consider this a lot more."

Several attendees, though, noted that it's difficult to ensure a high level of quality in data submitted by multiple sources to a shared resource and that this could be a potential stumbling block to the utility of a new database. For example, professor of pharmacy Robert S. Pearlman of the University of Texas, Austin, recounted concerns expressed by others that unintentional errors in submitted data, and even intentional ones, could undermine a grand-challenge type of chemical database.

"Maybe chemists are slow to the table compared to biologists in working together," Joyce said. "But if you're going to have this common ground, you have to talk about standards. It would behoove the field to get ahead of the curve on data quality standards, data formatting issues, and so on. Once you have that, then you won't have naysayers saying this is just a collection of garbage. But will pharma [the pharmaceutical industry] play? It will certainly be willing to take" information from a grand challenge database to aid its drug discovery program, "but will it also give?"

Trost said that "information should be as broadly available to the entire community as possible, but we also have to live in the real world. I certainly would hope that academic institutions would share their intellectual activities with virtually no restraints. Private companies that invest the huge amounts necessary to be successful in an activity like drug discovery must have some level of intellectual property protection, but one would hope that once they do obtain the necessary protection, they would indeed be willing to share. There, I think some companies are not as forthcoming as they could be and should be."

Associate research fellow Andrew Hopkins of Pfizer Sandwich Laboratories, in England, commented that "the pharmaceutical industry does publicly disclose much of its data in the form of patent applications and publications. However, there has been no commonly agreed standard or data repository for pharmacology and medicinal chemistry data, as there is in the sequencing and crystallography communities."

NIH is trying to encourage information sharing with its Road map for Medical Research, a planning framework introduced last September to round out and optimize the agency's research portfolio. The PubChem database is a road map component, and an NIH Chemical Genomics Center, announced earlier this month, is another.

The NIH center is the first component of what will be a nationwide consortium of chemical genomics centers. A library of up to 1 million chemical compounds will be established to support the centers. The centers will use high-throughput screening to "explore the vast majority of the human genome for which no small-molecule chemical probes have been identified," according to NIH, and the screening data will be deposited in PubChem. The ultimate goal is to enable new targets "to move more rapidly through the drug development pipeline." NIH has contracted with Kalypsys, San Diego, to provide the new NIH center with a high-throughput screening system, including automated robotics for running biochemical and cellular assays.

The chemical genomics center concept seems to directly address, at least in part, both of Schreiber's chemical grand challenge proposals--a shared database and the discovery of small-molecule modulators for every human protein function. John M. Schwab, program director for NIH's National Institute of General Medical Sciences, notes a distinction, however: "I don't think this really qualifies as an answer to the grand challenge to chemists, since this is an interdisciplinary resource, not one that's specific to chemists."

INVESTMENTS IN NEW TECHNOLOGIES. Also discussed was whether recent industry investments in advanced technologies have been worthwhile--such as the extensive investments that have been made in combinatorial chemistry, the synthesis and screening of large numbers of often related compounds.

Some attendees voiced concerns that such recent technology changeovers have for the most part been a washout. "The new fast route to drugs was supposed to start with big libraries, high-throughput screens, and all the great new genes that are being discovered," Joyce said. "But it hasn't played that way. Where are all the drugs?"

One drug industry researcher said his company invests in virtually every new scientific instrument system and technology under the sun, that many of them are then underutilized, and that the company would be far better off if it stopped this practice.

"Big pharma companies are in a lot of trouble," said chemistry professor JoAnne Stubbe of Massachusetts Institute of Technology. "They've spent all this money on combinatorial chemistry, and they haven't gotten nearly as many drugs in the pipeline as they needed to have. I think they need to be doing some serious thinking about why what they're doing is not working and if it's time to make some changes."

But others noted that some aspects of combinatorial chemistry are useful and are here to stay. "Some are a bit hard on combinatorial chemistry--almost too critical," Dobson said. "It's very important to have that sort of technology, and what we need to do now is look very hard at the results to understand what does work and what doesn't."

"There is a consensus in the industry that combinatorial chemistry has been less productive than anticipated 10 years ago," said Hans Ulrich Stilz, head of chemistry at Aventis Pharma Deutschland GmbH, in Frankfurt. "Like most technologies, combinatorial chemistry went through a certain amount of hype in the beginning. It went through a depression, and now it has come back to a level where it's considered to be a tool, like many other tools, that we have to use in an intelligent manner."

One problem with the use of combinatorial chemistry is that "we may have synthesized libraries without enough knowledge about what to synthesize to attract biological targets," Stilz said. "We now need to make better use of structural and functional knowledge to synthesize small, focused libraries of diverse structures." Work by several groups in the area of diversity-oriented synthesis exemplifies this trend.

Trost said that "companies seem never to be able to understand that there's no golden bullet that one can find to make drug discovery a much more efficient process. That doesn't mean you don't try to improve efficiency. But I am disappointed by how much companies will veer irrationally in one direction or another with the hope that they will improve efficiency. I just don't think that generating libraries of millions of compounds makes sense."

Nevertheless, "combinatorial chemistry clearly has a role to play," Trost added. "Small, focused libraries with truly more diverse structures could be useful. But natural products also have a very important role to play in drug discovery. People who say natural product research has had its day are being arbitrary and are limiting their options for no good reason."

DRUG DISCOVERY AND ACADEMIA. Should researchers in academia attempt to discover new drugs? Participants in general thought not, except perhaps for rare or Third World diseases.

Ansari questioned whether drug generation is a good use of the academic environment. "I think it raises a lot of issues," such as potential conflict-of-interest problems, he said. "What did come up at the meeting was the idea of greater collaboration between academics, nonprofit organizations, and the pharmaceutical industry to develop drugs for diseases that are rare and not economically feasible with the current business model."

Dobson said that "pressures on commercial organizations are such that they can increasingly tackle only the common diseases of rich countries. Finding a way to help those who suffer from pain and disease in the rest of the world represents another potential grand challenge. We all--whether in pharma, government, or academia--need to have innovative ideas on this issue."

According to Brent Stockwell, assistant professor of biological sciences and chemistry at Columbia University, "neglected diseases that the pharmaceutical industry won't tackle [include] orphan genetic diseases like ALS [amyotrophic lateral sclerosis], Huntington's disease, and spinal muscular atrophy, as well as Third World diseases that affect a large number of patients but don't represent a commercially valuable market, like malaria and schistosomiasis." Developing drugs for such diseases, he said, is "a significant challenge" and "a worthy goal."

But he noted that other attendees, "particularly from the pharmaceutical industry, had the somewhat perplexing perspective that this is not a worthy goal and should not be undertaken. Their primary argument was that drug discovery requires special expertise that really is only found in the pharmaceutical industry and that people who don't have that expertise should not try to discover drugs." Rather than criticize such academic drug discovery efforts, Stockwell said, "industry scientists should volunteer their valuable expertise to accelerate nonprofit efforts to tackle neglected diseases."

Stilz, commenting on the issue from the pharmaceutical industry side, said: "Academia can make a key contribution by laying a foundation for our sciences--by training scientists, exploring new chemistries and reactions that permit access to novel molecules, as well as exploring new biological mechanisms. As to whether academia should be involved in the discovery of drugs per se, I would be a little more cautious in my enthusiasm, because I do think that's applied research. It's nontrivial for academia to develop the infrastructure to do this effectively, and I would question whether it would be desirable."

Trost also said he believes that "academic institutions are not appropriate places for drug discovery" but that they nevertheless "have much to contribute in helping the drug discovery process. I hope that academic institutions will not lose sight of the fact that their primary mission is indeed education and begin crossing a line where the prime mission becomes whether they can make money or not. Academic institutions deservedly should be able to benefit economically from the intellectual property that's created there, but that shouldn't be the prime motivation."

Questions about drug industry practices in the hiring of chemists also arose at the meeting. Sir Jack E. Baldwin of the University of Oxford commented that big pharma has not, in general, sought a variety of backgrounds, viewpoints, and approaches in the chemists it hires. Others concurred, noting that the industry is known to strongly favor hiring of synthetic chemists instead of people trained more broadly in other chemistry subdisciplines.

Stilz said he agreed "with the view that it's good to bring in a diversity of talents. It's appropriate to strive for the right balance between skills and to hire scientists with problem-solving capabilities and broad backgrounds."

RUNNING OUT OF TARGETS? Horizon attendees also discussed why the pharmaceutical industry's drug pipelines recently have been relatively dry.

"You couldn't miss the tremendous pessimism that's out there in pharma," Joyce said. With a relatively small number of New Drug Applications in the hopper right now at the Food & Drug Administration, he remarked, "such pessimism is understandable." However, opinions on the root cause of the dry-pipeline problem varied quite a bit.

Drug industry researchers at the meeting pointed out that only about 3,000 human proteins are "druggable" (capable of being modulated by oral, small-molecule drugs); that only a fraction of these are likely to be useful targets for safe and efficacious drugs; that only about 200 of them are currently targeted by approved drugs; and that many of the others are difficult to reach with new drugs.

Another attendee predicted that 10 times the current number of known targets should be revealed fairly soon by proteomics research and studies of the recently determined human genome sequence. And Schreiber said he "would not be surprised if we eventually come to learn that there are as many as 20,000 druggable targets." However, Hopkins noted that such numbers may be unrealistically high because most sequence variations in proteins of the same type "do not affect drug binding sites and are thus practically indistinguishable by small-molecule drugs."

Complaining about the limited number of druggable targets is like "Chicken Little running around asking if the sky's falling," Trost said, "and I don't think the sky is falling. The large number of biotech company start-ups is a clear indication that the sky is the limit in terms of targets. This is almost a golden era in terms of what might be targets" because of the availability of the human genome sequence.

Hopkins replied that he and other attendees from industry "felt throughout the meeting that academics do not fully appreciate industry's desire to deliver drugs for unmet medical needs in a cost effective manner." Concentrating on a limited number of druggable targets enables company researchers "to focus on the 'high value real estate' in the genome," he said, instead of diluting their efforts by pursuing intractable targets.

Nevertheless, "you get what you look for, and I just think drug companies are not looking at things in the right way," Stubbe said. "That's what this whole meeting was about: Are we in the right chemical space? Are we using the right sets of assays? And I would say no." Drug companies think they have too few targets "because they have blinders on, in some fashion."

Companies need to change how they look for new targets, she said. "Over and over, you heard the same thing--new drug candidates must conform to Lipinski's rule of five. Maybe it's time to change Lipinski's rule of five." The rule of five, developed by Christopher A. Lipinski (now retired) and colleagues at Pfizer, Groton, Conn., is a set of criteria used to predict oral bioavailability from structure data. It's called the rule of five because the number five (or a multiple) appears in each of the rules, of which there are actually only four--involving molecular weight, number of hydrogen bond donors and acceptors, and lipophilicity.

Ansari agreed that "the reason things seem a bit dire for big pharma is because it has a very fixed paradigm. As we go forward, we will have to change that paradigm and think of things that go beyond Lipinski's rule of five."

However, Lipinski said he still believes that "oral activity is the current best option to help patients. It surprises me that people would question the drug industry's efforts to deliver oral drugs to patients in the most cost-effective manner possible."

In any case, Schwab believes that it's not the rule of five per se that is the major constraint to the discovery of new drugs. "Rather, it's the anticipated market," he said. "What I've learned is that pharma is focused almost single-mindedly on blockbuster drugs, and this appears to be driven largely by shareholder and management expectations. The number of targets for blockbuster drugs may be limited," he said, and this may be the basis of claims that there are relatively few druggable targets and that all the low-hanging fruit is gone--that the easily identified drugs have already been found. "Perhaps the NIH road map can serve as a stepladder to bring the rest of the fruit significantly closer."

Overall, the symposium had "a very ambitious agenda and some complex moments," Dobson concluded. "But it ended on a high point. People were recognizing the value of bringing chemistry and biology together, the complementarity of different approaches." The meeting also helped attendees better appreciate "that there really are some serious grand challenges that chemists can address that can have real impact on our own subject--as well as on the field of biology and the pharmaceutical industry."

Key questions, background articles, highlights, and summaries from this meeting, and from other Horizon Symposia, are posted at http://www.nature.com/horizon.